Health risk behavior among chronically ill adolescents: a systematic review of assessment tools

Research focusing on health risk behaviors (HRB) among adolescents living with chronic illness has increased over the past few decades. HRB are defined as specific forms of behavior associated with increased susceptibility to a specific disease or ill health on the basis of epidemiological or social data [1]. Examples of HRB include: alcohol, tobacco and drug use, unhealthy dietary habits, sexual behaviors contributing to unintended pregnancy and sexually transmitted diseases, behavior that contributes to unintentional injury or violence, and inadequate physical activity [2, 3]. In the past, it was presumed that chronically ill adolescents are restricted by their ill health from engaging in HRB [4, 5]. However, a growing body of evidence shows that chronically ill adolescents engage in such behavior at rates equivalent to [6–8] or at times higher [9–12] than their healthy peers. Some studies for example report higher frequency of cigarette smoking among adolescents with asthma [13, 14] and more substance or drug use among adolescents with mental illnesses [9, 15] compared to their healthy peers. In addition, chronically ill adolescents are often victims of behaviors resulting in unintentional injury and violence, such as bullying and sexual assault [16, 17]. Other problematic forms of HRB among chronically ill adolescents include; inadequate physical activity [18–20], risky sexual behavior [10, 11], and poor dietary habits [21].

Engagement in HRB is problematic for chronically ill adolescents because it hinders optimal care and management of ill health [22]. For example, studies among young people living with HIV report that anti-retroviral therapy adherence rates are poorer among the patients with riskier health lifestyle as compared to their HIV infected peers who have healthier lifestyles [23, 24]. Similarly, engagement in HRB such as tobacco use, recreational drugs use, and risky sexual behavior has been shown to hamper proper management of type 1 diabetes [25], asthma [26], and mental illness [27] among adolescents. Poor disease management compounded by direct adverse effects resulting from engagement in HRB, most likely translates into poorer health outcomes among chronically ill adolescents [5, 28]. Thus, promotion and maintenance of healthier behavioral practices early in adolescence has great potential to enhance positive long-term health outcomes for these patients [23].

Regarding the public health burden posed by HRB, frequent monitoring of such behaviors is recommended for supporting clinical and preventive efforts directed at improving lives of young people with chronic illnesses and their families [5, 29]. Although there are numerous measures of HRB, evidence is still meagre on the most frequently utilized HRB measures as well as the psychometric properties of HRB tools among chronically ill adolescents in various geographical contexts. Moreover, without proper adaptation, measurement bias and compromise to various psychometric properties like validity and reliability may arise [30, 31]. Bias also arises from unfamiliar content of the tests, translation challenges and unfamiliar means of tool administration [30]. Studies have similarly shown that variations in how questions are administered and how respondents are contacted affects the accuracy and quality of data collected [32]. There is still a lack of knowledge concerning the major forms of HRB, their commonly utilized assessment tools, their psychometric properties and their methods of administration in studies among chronically ill adolescents.

We therefore carried out this review to determine the current gaps in knowledge about tools to measure HRB. The review synthesizes findings from empirical studies conducted globally among adolescents living with chronic illnesses so as to: (i) identify the commonly utilized HRB assessment tools or sources of items used; (ii) describe the geographical utility of HRB assessments tools; (iii) identify the common means of HRB tool administration; (iv) document the reported adaptation and psychometric properties of HRB assessment tools or items; and (v) summarize the commonly assessed forms of HRB. We expect the results of this systematic review to aid HRB tool adaptation and validation procedures as well as enhance planning of research and interventions targeting adolescents living with chronic illnesses especially in low and middle income settings.

This systematic review was conducted following recommended guidelines for conducting systematic reviews [33]. We searched for relevant literature in four bibliographic databases: PubMed, Embase, PsycINFO and Applied Social Sciences Index and Abstracts. The search was initially conducted between November and December 31, 2015 and later updated in May 2017. The search strategy was formulated by two reviewers (DS and AA) and comprised of the following non-MeSH terms combined with Boolean operators: risk behavior OR risk taking OR health behavior OR healthy lifestyle AND adolescents OR Youth OR Teens AND Chronic condition OR Chronic disease OR Chronic illness. Additionally, other relevant studies were identified by searching the reference lists of the retrieved articles.

In this review, our study inclusion criteria were: (i) empirical studies published in a peer reviewed journal from January 1, 1980 to April 30, 2017; (ii) studies with participants aged 10–17 years or with mean age within this age bracket; and (iii) studies assessing for both HRB and chronic illness among the same study participants. The chronic conditions considered are those documented by the United States Department of Health and Human Services for the standard classification scheme [34]. Only studies published in English were included in this review. Studies were excluded if: (i) they were non-empirical (such as reviews, commentaries, letters to editor, conference abstracts), (ii) their participants had an age range or mean age below or above the 10–17 years’ category and (iii) they assessed only HRB without consideration of chronic illness or vise-versa.

Data extraction was done by two independent reviewers (DS, MKN). The data was extracted to Microsoft Excel spread sheets with the following details from eligible studies: author and date of publication, country where the study was conducted, age of the participants (mean age), form of chronic illness, assessment tool or source of items on HRB, methods of administration of HRB measures, psychometric properties of the tool (if documented), and form of HRB assessed were extracted. For reliability, we extracted measures of internal consistency, and interrater reliability such as the Cronbach’s alpha, intra-class coefficient (ICC) and coefficient of correlation whenever reported. For tool validity, we extracted construct, criterion, divergent or convergent validities whenever reported. We also noted any aspects of tool adaptation such as cultural adaptation, content validity, forward-back translations in case they were reported (refer to Table 4).

Data analysis involved collating and summarizing of results. The synthesis of data extracted from the eligible studies was done narratively. Frequencies and/or percentages were computed in Microsoft Excel program so as to summarize the findings on: the frequency of the various HRB tools/measures reported in studies, geographical utilization of these tools, forms of HRB assessed, methods of HRB tool/item administration and the various chronic conditions reported. Due to the high variation in HRB tools or items used, the tools were classified into four categories namely: (i) full version HRB assessment tools; (ii) modified version of HRB assessment tools; (iii) borrowed items on HRB; and (iv) items on HRB either newly developed or whose source is not specified by the author. Also in situations where more than one eligible manuscript was written using data from the same study, frequencies on HRB tools were collated in order to represent a single frequency count for this reported HRB assessment tool. For purposes of data management the reported chronic conditions were re-categorized into: respiratory, cardio-vascular, metabolic, hematological, mental, musculoskeletal, neurologic, dermatologic, digestive, physical disability and HIV.

The literature search yielded a total of 1623 articles and following a systematic appraisal of this literature (refer to Fig. 1), a total of 79 full articles were eligible for inclusion in this review.

Majority of the eligible studies were conducted in North America (60%) and Europe (24%). The rest of them were from Asia (8%), South America (2%), Oceania (2%) and a few were multi-site studies conducted in both Europe and North America (2%). The study site of one eligible study was not reported in the article [35].

The HBSC tool is a self-completion questionnaire administered in class room settings to adolescents aged 11–15 years and the HBSC study is conducted every 4 years across 44 countries in Europe and North America since its inception in 1982 [3]. The key health behaviors captured by this tool include; bullying and fighting, oral hygiene, physical activity and sedentary behavior, sexual behavior, substance use (e.g. alcohol, tobacco and cannabis), weight reduction behavior, behaviors resulting in injury, and dietary habits [3]. The YRBS tool (Standard and National High School questionnaires) is developed by the US Centers for Disease Control and Prevention (CDC) to monitor HRB that are considered leading causes of disability, death and social problems among youths in 9th to 12th grade (approximately 14–18 years) in the US Students complete the self-administered questionnaire during one class period and record their responses directly in an answer sheet. This tool assesses 6 forms of HRB: sexual risk behaviors, tobacco use, alcohol and other drug use, inadequate physical activity and unhealthy dietary behaviors [2].

The HRB tools were largely used among adolescents with the chronic conditions of mental illness, especially depression (21.4%), respiratory conditions such as asthma and cystic fibrosis (13.8%), metabolic conditions such as diabetes (9.4%) and neurological conditions such as autism spectrum disorders, epilepsy and cerebral palsy (6.9%). To a lesser extent, the HRB tools were also utilized among adolescent patients with musculoskeletal conditions such as arthritis, cardio vascular conditions (e.g. congenital heart disease and hypertension), HIV, cancer, digestive tract conditions (e.g. inflammatory bowel disease and gastritis), disabling conditions (e.g. visual, speech and hearing problems) and dermatological conditions such as atopic dermatitis and eczema. The detailed summary of eligible studies is presented in Table 4.

This review identified the commonly utilized HRB assessment tools or sources of items used; describing the geographical utility of HRB assessments tools, the common methods of HRB tool administration, the adaptation and psychometric properties; and providing a summary of the forms of HRB commonly assessed. Our findings show that the YRBS and HBSC are the most frequently used tools to assess HRB or sources of items on HRB. This may partly be explained by their high level of comprehensiveness in assessing priority and multiple forms of HRB thereby being useful in many contexts. While both tools assess for HRB among adolescents, the YRBSS targets an older adolescent age group compared to the HBSC. The HBSC however focuses more on the social and environmental context for HRB such as influence of peers, school environment, and family characteristics. The YRBSS explores HRB in greater detail compared to the HBSC although the former lacks items on oral hygiene, health complaints and chronic illnesses. Besides the YRBSS and HBSC, a wide range of other HRB tools have been utilized, and some of them assess the same form of HRB but in a different format. One challenge that this may present is the lack of uniformity or standardized formats to compare similar HRB outcomes across different study populations.

Findings from this review also indicate that research on HRB among adolescents living with chronic illnesses in low and middle income countries (LMIC) is still limited. This is unfortunate since the majority of the adolescent population lives in LMICs [56] where a disproportionately higher burden of HRB occurrence is also reported [57]. There are three potential reasons that may explain the limited research on HRB among chronically ill adolescents in LMICs. First there is limited research that explicitly focuses on the adolescent age-group [5]. Second, research on this topic is not adequately prioritized [4]. Nonetheless, research on HRB among chronically ill adolescents has significantly grown over the past two decades [4, 5] though with disproportionately lower prioritization especially in LMICs. The third reason is the scarcity of standardized measures on various health outcomes among chronically ill adolescents [5]. The need for more investment in research on health and behavioral outcomes among chronically ill adolescents especially in LMICs cannot be overemphasized given that the burden of chronic diseases is increasing in such settings [58].

The use of appropriate and psychometrically sound instruments is essential for having good insight in adolescents’ behavior so as to be able to address certain forms of behavior that could be dangerous either for the patients themselves or for others. However, our findings indicate that HRB tool adaptation and psychometric properties are rarely reported among studies on HRB of chronically ill adolescents. Partly, this could be due to the fact that the majority of the studies were conducted in the western context where the majority of these tools have been developed. To indicate the adaptation and psychometric properties, some of the authors simply cited studies where similar HRB tools or items have been previously utilized [59–61]. This may not guarantee validity and reliability for a number of reasons. First, some of the tools were previously adapted and validated for use among adolescents without chronic conditions and thus we cannot ascertain if they retain their good psychometric properties when used among chronically ill adolescents. Secondly, some of the original validation or adaptation may have taken place more than two decades back and considering the evolution of HRB, various behavioral constructs used in these tools may no longer be appropriate. Another observation is that many researchers borrow specific items from previously well validated or standardized HRB tools but without checking the item specific psychometric properties. Our findings also reveal that there is a tendency for researchers to perform the adaptation processes such as forward-back translation and content review for item completeness, clarity or cultural appropriateness; without performing psychometric evaluations. It should be emphasized that much as adaptation is an important process, psychometric evaluation is equally critical for ascertaining item reliability and validity. Without adequate adaptation and psychometric evaluation we cannot ascertain if the scales and items retain their good psychometric properties following the modifications made. Overcoming such challenges requires a mixed methods approach for tool adaptation and validation [31, 62, 63]. For instance, a four step approach has been suggested as adequate for adapting tools in low and middle income countries [64]. The four step approach suggested for LMICs entails: (i) construct definition which can be done through review of literature, and consultation with community or local professionals in order to achieve conceptual clarity and equivalence; (ii) item pool creation which involves preparation of a list of potentially acceptable items in a clear and unambiguous language using feedback from the first step; (iii) developing clear guidelines for administration of the items to ensure operational equivalence; (iv) test evaluation which involves psychometric evaluation to assess measurement and functional equivalence [64].

Additionally, findings from this review indicate that there are numerous methods of HRB tool or item administration. Self-administered paper and pencil format was the most popular method and this could have been because of the participants’ good level of literacy given that majority of them were school attending adolescents. This method of administration is also preferred as it is associated with a high level of privacy and ease of administration [32]. On the contrary, its disadvantage arises from its requirement for some literacy levels among the respondents as well as the cognitive burden that respondents face in comprehending and recalling their experiences [32, 65]. Face-to-face interviews were also frequently utilized in assessing HRB. This method is linked to high response rates and the benefit of probing participants and clarifying unclear questions [65]. Nonetheless, face-to-face interviews are hampered by the lack of anonymity which may result to social desirability bias and impression management [32, 65]. Similar to findings from other studies [32, 66], our review shows that there is growing utilization of electronic methods of HRB tool and item administration. Electronic methods [such as the Audio Computer Assisted Self Interview (ACASI), telephone and internet based surveys] are valued for their high level of privacy or anonymity [32, 65] and some of them such as the ACASI have been further designed to benefit people with low literacy levels [65]. However, electronic methods require access to electronic devices and services (such as telephone, computer, and internet), may require greater auditory demands and some demand a high level of literacy [32, 66]. The presence of numerous HRB tool administration methods presents a wide set of options which can be tailored to suit contextual factors, research skills, resource availability and specific needs of study populations. However, researchers should carefully think through the dynamics surrounding tool administration and data collection procedures in order to identify the most appropriate methods to ensure that high quality data is collected.

Furthermore, our findings show that alcohol, tobacco, drug use behavior and physical inactivity are the most frequently researched HRB among adolescents with chronic conditions. Substance use among chronically ill adolescents is of major concern and many studies report higher or equivalent rates of substance use (e.g. cannabis, tobacco, illicit drugs) among these adolescents in comparison to their healthy peers [12, 13, 67]. This may explain why most of HRB research among this group focuses on substance use behavior. Our findings also indicate that physical inactivity and sexual risk behavior are frequently assessed. Growing research interest on sexuality of chronically ill adolescents indicates that sexual risk behavior is a concern [7, 10–12] and this dissents the earlier notion that they are less sexually active than their healthy peers [4]. Likewise, physical activity among adolescents with chronic conditions is gaining measureable research interest [28]. This may surround its vital role in appropriate management of chronic illness such as: cardio-respiratory fitness among asthmatic patients and optimization of quality of life among patients with cerebral palsy [28]. Our results also indicate that violence related behaviors are frequently investigated among chronically ill adolescents. Adolescents with chronic illnesses often fall victim of violence such as bullying, assault and forced sexual encounters [17, 50]; and thus raising the need for increased research on this matter. On the other hand, our findings show that poor hygiene, inadequate sleep and behavior resulting to unintentional injury were the least frequently assessed forms of HRB in this review. This may be due to the reality that most of these problematic behaviors are of greater research interest in LMICs (whose representation is still low) where their occurrence is documented to be greater, compared to high income settings [57, 68]. Our findings on the variation in the frequency of the forms of HRB assessed, may partly imply that there is some tendency to measure HRB in isolation. However, co-occurrence of different adolescent HRB is increasingly documented [69, 70], and therefore different forms of HRB should be assessed concurrently.

Our review draws its major strengths from the utilization of a rigorous methodological framework [33] and also its specific focus on the adolescent age-group in a global perspective. However, we did not appraise the quality of the studies included in our systematic review. Nonetheless, given that our study objectives aimed at describing extent of utilization of HRB tools and providing an over-view of various forms of HRB assessed, we do not expect any major issues arising from the quality of studies to influence our findings.

Overall, most research on health risk behavior among chronically ill adolescents emanates from high income settings such as Europe and North America where the majority of the HRB assessment tools have also been developed. Therefore more investment is needed in research on health and behavioral outcomes among chronically ill adolescents especially in LMICs. Although the YRBSS and HBSC are utilized most, a variety of other HRB tools are used as well, however without documentation of adaptation and psychometric qualities. This poses challenges for researchers and practitioners who are keen to evaluate HRB in LMICs. We recommend the use of the mixed methods approach for tool adaptation and validation, which involves both qualitative approaches (e.g. focus group discussions and in-depth interviews) and quantitative approaches (e.g. psychometric testing) to develop and standardize measures for use by health researchers especially from LMICs. In the industrialized setting, we recommend the use of YRBSS or HBSC owing to their comprehensive approach to assessing multiple forms of HRB. The results of more research on HRB among chronically ill adolescents could translate to significant clinical, public health and social economic benefits, especially for adolescents living with such illnesses and their families.